CentOS Stream 10 vs. AlmaLinux 10 Beta vs. RHEL 10 Beta Performance Benchmarks

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • phoronix
    Administrator
    • Jan 2007
    • 67356

    CentOS Stream 10 vs. AlmaLinux 10 Beta vs. RHEL 10 Beta Performance Benchmarks

    Phoronix: CentOS Stream 10 vs. AlmaLinux 10 Beta vs. RHEL 10 Beta Performance Benchmarks

    Following the benchmarks earlier this month looking at the Red Hat Enterprise Linux 10 beta performance as well as the AlmaLinux 10 beta, on the same AMD EPYC server here are benchmarks when adding in CentOS Stream 10 to the mix. CentOS Stream 10 as the upstream to Red Hat Enterprise Linux 10 is largely similar to what's found in the RHEL 10.0 beta but one of the key differences is being powered by Linux 6.12 LTS rather than Linux 6.11 as currently used by the AlmaLinux/RHEL 10 beta. Here is how the performance of CentOS Stream 10 is looking in comparison on the same hardware.

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite
  • sophisticles
    Senior Member
    • Dec 2015
    • 2591

    #2
    No surprises here, there's no reason why all three, CentOS Stream 10, AlmaLinux 10 Beta and RHEL 10 Beta shouldn't perform within margins of error of each other.

    And the RHEL 9.5 results are due to using a much older kernel.

    I guarantee that if you shoehorn either 6.11 or 6.12 LTS into RHEL 9.5 you would find the performance to be flat between all four.

    I see no reason why anyone running RHEL 9.5 would bother with any of these three candidates and not just upgrade the kernel and call it a day.

    Comment

    • andyprough
      Senior Member
      • Feb 2012
      • 2452

      #3
      CentOS Stream is literally a meme distro, so its results are irrelevent.

      What's interesting is that the only difference between IBM-EL beta and Alma beta is statistical noise, so Alma is having no trouble keeping up with IBMHat step-for-step.

      Originally posted by sophisticles View Post
      No surprises here, there's no reason why all three, CentOS Stream 10, AlmaLinux 10 Beta and RHEL 10 Beta shouldn't perform within margins of error of each other.

      And the RHEL 9.5 results are due to using a much older kernel.​
      It's the grotesque mitigations against non-existent exploits that's likely holding 9.5 back - the newer kernels have some small bits of the grotesqueness scraped off.

      Comment

      • sophisticles
        Senior Member
        • Dec 2015
        • 2591

        #4
        Originally posted by andyprough View Post
        What's interesting is that the only difference between IBM-EL beta and Alma beta is statistical noise, so Alma is having no trouble keeping up with IBMHat step-for-step.
        Why wouldn't this be the case?

        Alma is just a rebranded RHEL, it would be odd if there was any significant performance difference.

        Comment

        • Radtraveller
          Phoronix Member
          • Mar 2020
          • 110

          #5
          So I was curious to run same tests on my lowly server to see.. and try to use the phoronix-test-suite .. but I am wondering how to run the exact same tests? I was looking through the article for maybe a command line text showing what and how your tests got run.. I cannot find.. So how can I duplicate the articles test?

          Comment

          • lejeczek
            Senior Member
            • Jul 2013
            • 126

            #6
            That lead of 10.x over 9.x is due to - my guess - kernel + relevant/crucial bits, being upped for that new CPU arch but, how would those performance gains (if any) look on an "older" CPU arch ? would be my question.

            Comment

            • adriansev
              Junior Member
              • Oct 2020
              • 22

              #7
              It would be really useful if any benchmark of EL type of OSes to be done also with ElRepo mainline kernel, as this can also give hints and information if the performance differences are kernel level or toolchain level .. IMHO for me the benchmarks presented in this article are irrelevant as the kernel baseline (latest mainline) is not taken into account.

              Comment

              Working...
              X